Professional Documents
Culture Documents
Andi Merxhani 1
am3232@caa.columbia.edu
This study is an introduction to the theory of poroelasticity expressed in terms of Biot’s theory of three-
dimensional consolidation. The point of departure in the description are the basic equations of elasticity (i.e.
constitutive law, equations of equilibrium in terms of stresses, and the definition of strain), together with the
principle of effective stress, and Darcy’s law of fluid flow in porous media. These equations, together with the
principle of mass conservation, are the only premises used to derive Verruijt’s formulation of poroelasticity as
used in soil mechanics. The equation of fluid mass balance derived in this work is an extension to Verruijt’s
original derivation, since it also considers the effect of the unjacketed pore compressibility (i.e. it accounts for
solid phase not being composed of a single constituent, and for the existence of occluded voids and/or cracks
within the solid skeleton.) Verruijt’s formulation uses a drained description where pore pressure is an inde-
pendent variable. Next, the increment of fluid content is defined and its constitutive law is derived - with its
derivation following naturally from the equation of fluid mass balance. Pore pressure, storage, and undrained
poroelastic coefficients are also introduced and useful relations are proven. Where appropriate, the physical
meaning of these coefficients is proven mathematically. Equations of equilibrium and fluid mass conservation
are subsequently expressed in terms of the increment of fluid content and undrained coefficients, leading to an
undrained description of poroelasticity. Thus Verruijt’s approach is extended to Rice and Cleary’s formalism.
This approach to poroelasticity is useful for its simplicity. It does not require the ad-hoc definition of poroelastic
constants and that of an elastic energy potential. Instead, it is a direct extension to isothermal linear elasticity
that accounts for the coupling of skeletal deformations and fluid behaviour.
6 Fluid diffusion 21
7 Storage coefficients 22
7.1 Physical interpretation of storativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
7.2 Uniaxial storage coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.2.1 Diffusion equation in one-dimensional consolidation . . . . . . . . . . . . . . . . . . 23
10 Epilogue 36
2
List of notations
3
Eu undrained Young’s modulus, (Eqn. 135)
G shear modulus, (Eqn. 1)
K bulk modulus, (Eqn. 1)
Kuundrained bulk modulus, (Eqns. 125, 132, 136)
Kfbulk modulus of the fluid phase
Ksbulk modulus of the solid phase
M Biot’s modulus, (Eqn. 94, 126)
S uniaxial storage coefficient, (Eqns. 109, 110, 113, 114)
Sǫ storage coefficient, (Eqns. 86, 87, 120)
Sr degree of saturation
4
1 Introduction
A description of the mechanical behaviour of fluid saturated porous media under the assumption of small per-
turbations is presented. The treatment falls within the framework of Biot’s theory of consolidation, thus it is
phenomenological. The instigator for the development of poroelasticity was the solution to the problem of
soil consolidation 2 . The first treatment of this problem was a phenomenological approach by Terzaghi [1925,
1943], who considered soil to be laterally confined, thus undergoing uniaxial deformations. In Terzaghi’s ap-
proach both solid and fluid constituents of the porous medium are considered incompressible. A general three-
dimensional theory of elastic deformation of fluid infiltrated porous media was proposed by Biot [1941], in
which the limitation of incompressible constituents was removed. Furthermore, the increment of fluid content
per unit volume was introduced as a variable work conjugate to the pore pressure. In Biot [1955], the theory
was extended to the general anisotropic elastic case. The equations for the dynamic response of porous media
were derived in Biot [1956], while extensions to nonlinear elasticity were presented in Biot [1973]. A formu-
lation of Biot’s linear theory suitable for problems of soil mechanics was proposed by Verruijt [1969], while
Rice and Cleary [1976] reformulated the equations of consolidation in terms of undrained coefficients. Thus,
the distinction between drained and undrained description of the equations of consolidation was introduced.
Extensions to the use of nonlinear constitutive law were proposed among others by Zienkiewicz et al. [1980]
and Prevost [1980, 1982]. A general treatment of Biot’s theory in the range of nonlinear material behaviour and
large deformations can be found in Coussy [1991, 2004], where the theory is reformulated using a thermody-
namics approach.
Fluid infiltrated porous media consist of solid skeleton and fluid material that occupies the porous space.
The mechanical behaviour of such media accounts for the coupling of skeletal deformations and fluid behaviour.
The material considered here is isotropic, and undergoes quasi-static deformations under isothermal conditions.
Porosity refers only to the connected porous space - however, the existence of isolated voids or cracks within
the solid skeleton is not excluded. Solid phase is compressible and is not necessarily composed of a single
constituent 3 . Pore fluid is compressible and consists of a single phase - for example, it can be water containing
isolated air bubbles. The range of applicability of this theory is typically for a degree of liquid saturation higher
than 90%. For lower degrees of saturation in the range of 0.75 − 0.8, researchers have used Biot’s theory still
assuming that the air and liquid pressures are equal (see, for example, Okusa [1985]).
A fundamental principle used in the description of the mechanical behaviour of porous media is the prin-
ciple of effective stress. This principle describes the decomposition of internal stresses applied to a porous
medium. Accordingly, part of the stresses applied are transmitted to the pore fluid and the rest are transmitted
to the solid skeleton. The former component causes changes in pore pressure and subsequent fluid flow. The
latter component is the effective part of stresses that causes deformations on the solid skeleton. Consequently,
the equations of equilibrium are the same as in classical elasticity, only expressed in terms of effective stresses.
As it is reasonable to expect, conservation of mass is also required for the mechanical description to account
for fluid flow. Fluid flow is considered to be viscous and is governed by Darcy’s law. The fluid phase manifests
itself in the equations through pore pressure, p, or the increment of fluid content per unit volume, ζ. The use of
the increment of fluid content in the equations of equilibrium is associated with undrained poroelastic constants
and leads to the undrained description of the equations of poroelasticity.
2 Excellent reviews of the initial developments in the theory of porous media have been written by de Boer [1996, 2000]
3 The effect of voids and cracks occluded within the solid skeleton, and the existence of multiple solid constituents are included through
the consideration of the unjacketed pore compressibility in the storage and pore pressure coefficients.
5
The present text has been written having in mind readers aiming at a first introduction to the theory of
poroelasticity - particularly engineers with a background in soil mechanics who would like to access the general
literature in the field of poroelasticity. It aims at bridging the gap between the formulation of poroelasticity
as used in the field of soil mechanics, with the generality of the formulation of Rice and Cleary, favoured
in problems of rock mechanics. The exposition is phenomenological. The point of departure are the basic
equations of elasticity (i.e. constitutive law, equations of equilibrium in terms of stresses, and the definition
of strain), together with the principle of effective stress, from which the poroelastic equations of equilibrium
are derived. Next, the equation of fluid mass conservation is introduced as derived by Verruijt. The expression
derived herein is slightly more general, as it also includes the effect of the unjacketed pore compressibility.
So far pore pressure is used as an independent variable in the description. The only new coefficients that are
necessarily introduced in the equations are Biot’s coefficient that appears in the principle of effective stress, the
storage coefficient that appears in the equation of mass conservation, and definitions of material compressibility.
Next, the increment of fluid content is defined and its constitutive law is derived, following naturally from the
equation of fluid mass conservation. Pore pressure, storativity, and undrained poroelastic coefficients are also
introduced and useful relations are proven. Equilibrium and mass conservation are subsequently expressed in
terms of the increment of fluid content and undrained coefficients, leading to Rice and Cleary’s formalism.
Lastly, a weak (u, p) formulation of poroelastic problems is presented.
6
2 Basic equations of isotropic elasticity
Before proceeding to the exposition of the theory of consolidation, it is useful to first review the basic equations
of elasticity - both for compressible and incompressible elasticity. The equations and derivations related to this
section can be found, for example, in Westergaard [1952]. The equations of the classical theory of elasticity
can be fully defined using two material constants. An appropriate pair of elasticity constants can be Young’s
modulus, E, and Poisson’s ratio, ν. Other fundamental constants that can be used are the bulk modulus, K, the
shear modulus, G, and Lamé’s constant, λ, which are linked to Poisson’s ratio and Young’s modulus by the
relations
E E Eν
K= ; G= ; λ= (1)
3(1 − 2ν) 2(1 + ν) (1 + ν)(1 − 2ν)
Further useful relations among elasticity constants are the following:
2 2ν 2(1 + ν)
λ = K − G; λ=G ; K=G (2)
3 1 − 2ν 3(1 − 2ν)
Using index notation, the components of the strain tensor are defined as
1 ∂ui ∂uj
ǫij = + (3)
2 ∂xj ∂xi
where ui is the ith component of displacement. The constitutive law can be written in a strain-stress relation as
1+ν 3ν
ǫij = σij − σδij (4)
E E
where σ is the mean stress σ = σkk /3 and δij is Kronecker’s delta. Inverting (4) the stress-strain relations are
obtained as
2
σij = (K − G)ǫδij + 2Gǫij (5)
3
with ǫ being the volumetric strain, which is the first invariant of the strain tensor defined as ǫ = ∇ · u. From
the above equation the relation connecting mean stress to volumetric strain can be retrieved
σ = Kǫ (6)
∂σij
+ bi = 0 (7)
∂xj
∂ǫij ∂ǫ
2G +λ + bi = 0 (8)
∂xj ∂xi
7
Furthermore, ǫij is given in (3), which, substituted into equations (8) yields
∂ 2 ui
∂ ∂uj ∂ǫ
G +G +λ + bi = 0 (9)
∂xj ∂xj ∂xi ∂xj ∂xi
∂ 2 (·) ∂uj
∇2 (·) = ; ǫ= (10)
∂xj ∂xj ∂xj
3ν
σij = 2Gǫij + σδij , and ǫ = σ/K (12)
1+ν
in which the mean stress, σ, is viewed as an independent variable. Relation (12) can be obtained by substituting
(6) into (5). The difference in this case though, is that equations (12) are valid even for ν = 1/2, which results
to K → ∞. Substituting (12) into the equilibrium equations (7), yields
∂ǫij 3ν ∂σ
2G + + bi = 0 (13)
∂xj 1 + ν ∂xi
or
∂ 2 ui
∂ ∂uj 3ν ∂σ
G +G + + bi = 0 (14)
∂xj ∂xj ∂xi ∂xj 1 + ν ∂xi
which due to (10) become
∂ 3ν ∂σ
G∇2 ui + G ǫ+ + bi = 0, ǫ = σ/K (15)
∂xi 1 + ν ∂xi
In the limit of material incompressibility, where ν = 1/2, the bulk modulus tends to infinity (K → ∞). Since
stresses are finite, then relation ǫ = 0 should hold. Therefore, for incompressible material behaviour, and using
vector notation, the equations of equilibrium become
8
3 Principle of effective stress
Stresses applied to a saturated porous medium are partly distributed to the solid skeleton and partly to the
pore fluid. The former stresses are responsible for skeletal deformations, this is why they are called effective.
Considering that stresses are positive when they are tensile and pressure is positive when it is compressive, the
principle of effective stress is written in index notation as
′
σij = σij − αpδij (17)
′
In equation (17), σij and σij are the components of the total and effective stress and p is the pore pressure. The
symbol δij is Kronecker’s delta, defined as δij = 1 for i = j, and δij = 0 for i 6= j. The parameter α is known
as Biot’s coefficient. In the work of Terzaghi [1943], this parameter was considered to have the value of one -
an assumption generally valid for soil.
Biot’s coefficient α
Biot [1941] expressed coefficient α of equation (17) as
K
α= (18)
H
where K is the drained bulk modulus of the porous material, and 1/H is the poroelastic expansion coefficient,
that was introduced by Biot. It describes the change of the bulk volume due to a pore pressure change while the
stress is constant. Biot and Willis [1957] recast the above equation in terms of two coefficients of unjacketed
and jacketed compressibility. The unjacketed coefficient of compressibility is the compressibility of the solid
phase Cs 4 and the jacketed compressibility coefficient is the drained compressibility of the porous material C
5
. In familiar notation of soil mechanics the expression they provided is
Cs
α=1− (19)
C
The above relation has also been derived independently from Bishop and Skempton (Skempton [1960]). A
derivation can also be found in Bishop [1973].
For soft soils, the value of α is considered to be one and the principle of effective stress reduces to
′
σij = σij − pδij (20)
This is the expression used by Terzaghi in his original work and is valid for most soils. Soils are usually soft
and their skeleton is highly compressible, while their particles (solid phase) have small compressibility, which
4 The compressibility of the solid phase is often considered identical to the compressibility of the grains. This, however, is true if the
[1957]). To perform the test a sample is immersed in fluid under pressure, with the fluid penetrating the pores of the material. Any change
in the applied confining pressure produces an equal change to the pore pressure. Denote the confining pressure with ∆Pc and pore pressure
with ∆p, then for this experiment the condition ∆Pc = ∆p holds. The compressibility of the solid phase is calculated as
1 ∆V
Cs = − |∆Pc =∆p
V ∆p
The jacketed compressibility is measured in a drained test, where the sample is covered by a surface membrane and the inside of the
jacket is connected to the atmosphere using a tube. The tube connection to the atmosphere makes the pore pressure remain constant under
the load applied to the specimen.
9
justifies the use of the coefficient α as equal to 1. In contrast, this is not always the case for rocks. Table 1
presents typical values of Biot’s coefficient and is compiled with data obtained from Mitchell and Soga [2005].
1+ν ′ ν
ǫij = σ − σ ′ δij (21)
E ij E kk
′
Equations (21) differ from (4) in that stresses σij are the effective stresses applied on the soil skeleton. From
equations (21) the equivalent stiffness formulation can be derived as
′ 2G
σij = K− ǫδij + 2Gǫij (22)
3
where ǫ is the volumetric strain, ǫ = ǫkk (indices appearing twice indicate summation under Einstein’s conven-
tion). Making use of the principle of effective stress (equation 17), the stress-strain relations (21) and (22) are
expressed in terms of the total stresses and pore pressure as
1 ν α
ǫij = σij − σkk δij + pδij (23)
2G 1+ν 3K
10
and
2G
σij = K− ǫδij + 2Gǫij − αpδij (24)
3
From equation (22), the isotropic (or mean) effective stress can be expressed with respect to the bulk modulus
of the porous material and the volumetric strain as
′
σkk
= Kǫ (25)
3
which, in terms of the compressibility and mean effective stress, is expressed as
ǫ
σ′ = (26)
C
Making use of the principle of effective stress (17), equation (25) can be written in terms of mean stress as
σ + αp = Kǫ (27)
Last, stress-strain relationship with regard to mean stress can be derived by substituting the latter equation into
equation (24) and making use of the relations (2) thus reads
3ν αp
σij = 2Gǫij + σδij − 2G δij (28)
1+ν 3K
Therefore, the constitutive relation (24) can be substituted by the two relations (28) and (27).
The term mv in equation (29) represents the drained (i.e. p = 0) vertical compressibility of the laterally con-
fined soil. The value of mv can be calculated using equations (21). Considering the boundary conditions of the
one dimensional consolidation problem which allow only for vertical frictionless displacements, shear strains
and horizontal strains are set to zero. Vanishing shear strains lead to shear stresses being zero. Furthermore,
substituting ǫx = 0 and ǫy = 0 into (21) yields
2ν 2 ′
ν(σx′ + σy′ ) = σ (30)
1−ν z
(1 + ν)(1 − 2ν) ′
ǫzz = σzz (31)
E(1 − v)
11
which makes the value of mv equal to
(1 + ν)(1 − 2ν)
mv = (32)
E(1 − ν)
or
1
mv = (33)
(λ + 2G)
In equations (32) and (33), E and ν are the Young’s modulus and the Poisson ratio of the porous medium after
the excess water is squeezed out (drained), respectively, G is the shear modulus, and λ is a Lamé constant
(drained).
σ = σ ′ − αmp (35)
(Zienkiewicz et al. [2005, 1999]), where m is the vector form of Kronecker’s delta, δij . In three dimensions m
reads
m = [1, 1, 1, 0, 0, 0]T (36)
σ ′ = Dǫ (37)
The matrix D in elasticity theory is often called the stiffness matrix and in three dimensions it is equal to
1−ν ν ν 0 0 0
ν 1−ν ν 0 0 0
ν ν 1−ν 0 0 0
E
1 − 2ν
D= (38)
(1 + ν)(1 − 2ν) 0 0 0 0 0
2
0 1 − 2ν
0 0 0 0
2
1 − 2ν
0 0 0 0 0
2
12
Lastly, strains and displacements are linked through the kinematics relations
ǫ = ∇s u (39)
with u = [u, v, w]T . ∇s appearing in the above equation denotes the symmetric gradient operator, which for
three dimensional problems is defined as
∂/∂x 0 0
0 ∂/∂y 0
0 0 ∂/∂z
∇s =
∂/∂y ∂/∂x
(40)
0
0 ∂/∂z ∂/∂y
∂/∂z 0 ∂/∂x
∇Ts σ + b = 0 (42)
or in index notation ′
∂σij ∂ρ
− αδij + bi = 0 (45)
∂xj ∂xi
The substitution of equation (41) in the last equation, leads to the equilibrium equations in term of displace-
ments
∇Ts D∇s u = ∇Ts (αmρ) − b (46)
Using index notation, the equations of equilibrium (46) take the form
E 1 ∂ǫ ∂p
∇2 ui + −α = −bi (47)
2(1 + ν) 1 − 2ν ∂xi ∂xi
with ǫ = ∂u ∂v ∂w
∂x + ∂y + ∂z being the volumetric strain in three dimensions. Introducing the shear modulus and
Lamé’s constant λ, as defined in equations (11 ) and (22 ), the equations of equilibrium (47) are written as
13
∂ǫ ∂p
G∇2 ui + (λ + G) =α − bi (48)
∂xi ∂xi
The system of equations (48) contains one variable in excess. An additional equation is required to complement
the boundary value problem, and it is obtained from the conservation of fluid mass, which is examined in section
4.5.
For the case of plane strain there is one kinematic compatibility equation, which is first derived in terms of
strain. The strain-displacement relations for plane strain are
∂u ∂w ∂u ∂w
ǫxx = ; ǫzz = ; γxz = + (49)
∂x ∂z ∂z ∂x
Eliminating the displacements from equations (49) leads to the strain compatibility equation
′ ′ ′
σyy = ν (σxx + σzz ) (51)
6 Which are the stresses and the pore pressure.
14
Furthermore the shear stresses on the x − y and z − y directions are zero, because the equivalent strains are
zero also. Substituting equation (51) to (211 ) and (212 ), the expressions for ǫxx and ǫzz become
1
(1 − ν 2 )σxx − (ν + ν 2 )σzz
′ ′
ǫxx =
E
(52)
1
(1 − ν 2 )σzz − (ν + ν 2 )σxx
′ ′
ǫzz =
E
′ ′
Next the shear stress σxz is expressed with respect to σxx , σzz , and p, using the equilibrium equations (43),
which for plane strain conditions are expanded as
′
∂σxx ∂σxz ∂p
+ −α = −bx
∂x ∂x ∂x
(53)
∂σxz ∂σ ′ ∂p
+ zz − α = −bz
∂z ∂z ∂z
Derivating the first of the equations (53) with respect to x, the second with respect to z, and adding them,
results to
∂ 2 σxz
2 ′
∂ 2 σzz
′
∂ σxx 2
2 =− + − α∇ p − ∇ · b (54)
∂x∂z ∂x2 ∂z 2
where b = (bx , bz )T and ∇ · b = ∂bx /∂x + ∂bz /∂z. Shear stress-strain relation is
2(1 + ν)
γxz = σxz
E
which under equation (54) becomes
The Beltrami-Michell equation for plane strain is now obtained substituting equations (52) and (55) to the strain
compatibility equation (50), which in terms of the effective streses reads
2′ ′ αp 1
∇ σxx + σzz − =− ∇·b (56)
1−ν 1−ν
1
∇2 (σxx + σzz + 2ηp) = − ∇·b (57)
1−ν
α(1 − 2ν)
η=
2(1 − ν)
Compatibility equation (4.3.1) associated with equations (53) and the equation of mass continuity, can be used
as a full set of equations to solve plane strain poroelastic problems. This formulation is suitable when only
stress boundary conditions are applied. It should be noted at this point that for multiply connected domains,
15
compatibility conditions are necessary but not sufficient condition to guarantee single-valued (continuous)
displacements. For a discussion on this topic please refer to Section 10.4 of Chou and Pagano [1992].
The equations of strain compatibility in three dimensional elasticity are derived in a process similar to the
derivation of equation (50) It is found that they form the system of six equations
1 ∂ 2 σkk ∂2p
′
2 ′ ν 2 ν ∂bi ∂bj
∇ σij + −α 2 + δij ∇ p = − δij ∇ · b − − (64)
1 + ν ∂xi ∂xj ∂xi ∂xj 1−ν 1−ν ∂xj ∂xi
where b = (bx , by , bz )T . A second form which equation (64) can take using total stresses, is
1 ∂ 2 σkk 1 − ν ∂2p
ν ∂bi ∂bj
∇2 σij + + 2η + δij ∇2 p = − δij ∇ · b − − (65)
1 + ν ∂xi ∂xj 1 + ν ∂xi ∂xj 1−ν ∂xj ∂xi
where the coefficient η is the same as in (4.3.1). Last, contracting equations (65) by setting i = j, results to the
useful equation
1+ν
∇2 (σkk + 4ηp) = − ∇·b (66)
1−ν
k
q=− ∇(p − ρf zg) (67)
γw
16
(Wang [2000]), where k is the specific permeability (or hydraulic conductivity), and γw is the specific weight
of the fluid, defined as γw = ρf g. The vector q is specific discharge, that is, the relative velocity of the fluid
component of a porous medium, vf , with respect to the velocity of the solid component, vs , multiplied by the
porosity, n, and is represented as
q = n(vf − vs ) (68)
∂n ∂p
+ nβ + ∇ · [n(vf − vs )] + ∇ · (nvs ) = 0 (72)
∂t ∂t
The term n(vf − vs ) in the above equation is identified with the specific discharge, q, while the term ∇ · (nvs )
is calculated using the conservation of mass of the solid skeleton.
Consider now the mass balance of the skeleton. Denoting the density with ρs , the solid mass is equal to
ms = (1 − n)ρs V , and the relative density is equal to (1 − n)ρs . Same as with equation (69), mass balance is
written as
∂[(1 − n)ρs ]
+ ∇ · [(1 − n)ρs vs ] = 0 (73)
∂t
7 This relation can be shown as follows. Consider constant fluid mass, m . Mass continuity dictates that ρ V
f f o f o = ρf Vf , where ρf o
and Vf o are fluid density and volume at reference configuration and ρf and Vf are the equivalent quantities in the current configuration.
∆Vf ∆ρf
Substituting for Vf = Vf o + ∆Vf and ρf = ρf o + ∆ρf into the above equation of fluid mass continuity yields =− . In
Vf o ρf
the range of small perturbations this can be written as
∂Vf ∂ρf
=−
Vf ρf
17
It now remains to derive the constitutive law for density of the solid phase. This constitutive law can
be derived considering the volumetric response of a porous element (in the following, equation 26a of
Detournay and Cheng [1993] is derived). Consider a linear elastic porous medium of porosity n, saturated
with fluid and loaded with an isotropic compressive stress of ∆Pc under undrained conditions. This loading
condition causes within the specimen mean total stress of magnitude
∆σ = −∆Pc
(tensile stresses are positive), and increase in pore pressure, ∆p. The difference between the confining load and
pore pressure is denoted with ∆Pd = ∆Pc − ∆p.
Within the framework of an imaginary experiment, consider that the load is applied in two stages. In the first
stage an increment of confining pressure ∆p, and equal change in pore pressure are applied (this is essentially
the unjacketed test as described in Section 3). Therefore at this stage ∆Pd = 0. In the second stage a confining
load of ∆Pc − ∆p is applied without any increase in the pore pressure, and ∆Pd = ∆Pc − ∆p.
Consider the first stage of loading (where ∆Pd = 0). It is convenient here to be reminded of the definition
of the compressibility of the solid phase, i.e. the unjacketed bulk compressibility
1 ∆V
Cs = − |∆Pd =0 (74)
V ∆p
Another useful definition is that of the unjacketed pore compressibility
1 ∆V p
Cφ = − |∆Pd =0 (75)
V p ∆p
where Vp = nV is the volume of pores. Compressibility Cφ is identified with 1/Ks′′ of Rice and Cleary [1976].
For saturated media this is equal to the volume of fluid phase. Compressibility Cφ is usually considered to
be equal to Cs . However, this is true for porous materials of which the solid phase is composed of a single
constituent - this requirement also excludes the presence of entrapped fluid and presence of cracks within the
solid skeleton. For more details refer to Wang [2000], Section 3.1.4. Total volume is composed of the volume
of the solid phase, Vs = (1 − n)V , and that of the pores, Vp = nV , such that V = Vc + Vp . The fractional
volume change of the solid phase can be written as
∆Vc ∆V ∆Vp
= − ,
Vc Vc Vc
or
∆Vc 1 ∆V n ∆Vp
= −
Vc 1−n V 1 − n Vp
∆V ∆Vp
Definitions (74) and (75) suggest substitutions = −Cs ∆p and = −Cφ ∆p into the above equation,
V Vp
yielding
18
∆Vc −Cs (∆Pc − ∆p)
= (77)
Vc 1−n
Combining the effects in equations (76) and (77), and making the substitution ∆σ = −∆Pc , the total
volume change of the solid phase is obtained as
∆Vc Cs ∆σ + nCφ ∆p
= (78)
Vc 1−n
(this is equation 26a of Detournay and Cheng [1993]). Departing from equation (78) and making the same
arguments made for the derivation of the constitutive law for the fluid phase (70), the constitutive law for the
solid phase can be derived as
∂ρs ρs ∂σ ∂p
= −Cs − nCφ (79)
∂t 1−n ∂t ∂t
Substituting (79) into (73) results to
∂n ∂σ ∂p
− + −Cs − nCφ + ∇ · vs = ∇ · (nvs ) (80)
∂t ∂t ∂t
The term ∇ · (nvs ) as calculated in equation (80) is substituted into the equation of fluid mass conservation
(72), that now reads
∂p ∂σ
∇ · vs + ∇ · [n(vf − vs )] + n(β − Cφ ) − Cs =0 (81)
∂t ∂t
∂ǫ
The quantity n(vf − vs ) is the specific discharge defined in (68), and = ∇ · vs is the time derivative of the
∂t
volumetric strain. Equation (81) now is written as
∂ǫ ∂p ∂σ
+ n(β − Cφ ) − Cs = −∇ · q (82)
∂t ∂t ∂t
Equation (82) can be further simplified by substituting for the divergence of the specific discharge. Furthermore
the mean (or isotropic) total stress σ can be split into the mean effective stress and pore pressure components.
The mean effective stress can be expressed with respect to the compressibility of the porous material and the
volumetric strain as
ǫ
σ′ = (83)
C
Finally, making use of (19), fluid conservation equation is written as
∂ǫ ∂p
α + Sǫ = −∇ · q (84)
∂t ∂t
In this form the equation of fluid mass conservation in fluid infiltrated porous media is known as the storage
equation (Verruijt [1969]). The term −∇ · q appearing in equation (84) is given by Darcy’s law (67), so that
equation (84) can be written as
∂ǫ ∂p k
α + Sǫ =∇· ∇p (85)
∂t ∂t γw
The quantity Sǫ appearing in the storage equation equals
19
and, in literature, this is known as the storativity of the pore space. As will be seen in subsequent section,
storativity can be expressed in terms of Skempton’s pore pressure coefficient and Biot’s coefficient (see equation
119). For practical purposes it is easier to calculate storativity by measuring Skempton’s coefficient and avoid
the difficulties of measuring Cφ .
For porous materials with solid phase composed of a single constituent, coefficients Cs and Cφ are equal
and Sǫ is expressed as
Sǫ = nβ + (α − n)Cs (87)
with β defined in (88) for the case of fluid-gas mixtures and Biot’s coefficient, α, considered to be 1.
where Uf and Us signify the average displacement of the fluid and solid phases in the control volume. Equation
(90) holds under the assumption that porosity does not vary in space. Alternatively, n should be put inside the
8 Rice and Cleary [1976] defined the increment of fluid content as follows. Consider the mass of fluid, M , contained in a control vol-
f
ume, V , of porous material. Define the fluid mass content per unit volume as m = Mf /V , where V is in an unstressed and unpressurised
state. Define also the apparent fluid volume fraction as υ = m/ρo , where ρo is the reference state mass density of fluid. The increment of
fluid content is simply
ζ = υ − υo
where υo is the reference value in the unstressed state.
20
parentheses in equation (90) (Wang [2000], Section 2.1.2). Taking the time derivative of the right hand side
terms of (90), results simply to the specific discharge velocity defined in equation (68). Considering equations
(90) and (68), in the absence of fluid sources, the following fluid continuity equation is found to hold
∂ζ
= −∇ · q (91)
∂t
Next consider the storage equation (84), written in the form
∂
(αǫ + Sǫ p) = −∇ · q (92)
∂t
From equations (84) and (91), and integrating with respect to time considering unstressed reference state, the
constitutive relation for the increment of fluid content is derived as
ζ = αǫ + Sǫ p (93)
1
M= (94)
Sǫ
the constitutive law of the increment of fluid content (93) takes the form
1
ζ = αǫ + p (95)
M
6 Fluid diffusion
The equation of mass conservation can be expressed in terms of the increment of fluid content, ζ, resulting in a
diffusion type equation. Specifically, when the storage equation (91) is considered with Darcy’s law (67), and
under the assumption that k/γw does not vary in space, it takes the form
∂ζ k 2
= ∇ p (96)
∂t γw
Next, the right hand side of (96) is expressed in terms of the increment of fluid content. The required expression
can be derived from the equations of equilibrium (48). Taking the derivative of the i − th equation of (48) with
respect to xi and summing the resulting three equations, in the absence of body forces yields
λ + 2G
∇2 p − ǫ =0 (97)
α
Substituting equation (93) into equation (97) the required expression is obtained as
λ + 2G
∇2 p = ∇2 ζ (98)
α2 + (λ + 2G)Sǫ
Combining equations (96) and (98), the equation of mass conservation takes the form of a diffusion equation
as
∂ζ
= c∇2 ζ (99)
∂t
21
Equation (97) used to derive the above equation was derived neglecting body forces. In the general case where
body forces and fluid source are present, equation (99) takes the form
∂ζ
= c∇2 ζ + αmv c∇ · b + Q (100)
∂t
where b is the vector of body forces, coefficient mv is as defined in (32), and Q is the fluid source term (for
the mathematical treatment of the source term please refer to Wang [2000], Rudnicki [1986], or Chau [2012]).
The coefficient, c, is the coefficient of consolidation in three dimensions, and is equal to
k λ + 2G
c= · (101)
γw α2 + (λ + 2G)Sǫ
or
k (λ + 2G)M
c= · (102)
γw α2 M + λ + 2G
where M is Biot’s modulus defined in equation (94).
7 Storage coefficients
Two storage coefficients are examined in this section, storativity and the uniaxial storage coefficient. Storativity,
Sǫ , was introduced in equation (84) of mass conservation and its physical meaning is shown in this section.
The uniaxial storage coefficient is the equivalent storage coefficient for laterally confined deformations and is
also introduced.
∂ǫ ∂p ∂ζ
α + Sǫ = (103)
∂t ∂t ∂t
Pore pressure and volumetric strain depend on time, while the increment of fluid content can be written as
ζ(p(t), ǫ(t)). Its partial derivative with respect to time is
∂ζ ∂ζ ∂p ∂ζ ∂ǫ
= + (104)
∂t ∂p ∂t ∂ǫ ∂t
For constant control volume the time derivative of volumetric strain vanishes from both equations (103) and
(104). Furthermore, substituting (104) to (103) yields
∂ζ ∂p
Sǫ − =0 (105)
∂p ∂t
The above equation involves multiplication of scalar quatities, from which the following relation is implied
∂ζ
Sǫ = (106)
∂p ǫ=0
22
Expression (106) gives to Sǫ the inerpretation of being the fluid volume change per unit control volume and
per unit pressure change, while the control volume remains constant.
∂ǫzz ∂p ∂ζ
α + Sǫ = (107)
∂t ∂t ∂t
′ ′
Using the one-dimensional constitutive law (ǫzz = mv σzz ) and the principle of effective stress (σzz = σzz +
αp), equation (107) becomes
∂σzz ∂p ∂ζ
αmv + (Sǫ + α2 mv ) = (108)
∂t ∂t ∂t
A new storage coefficient can now be defined as
S = Sǫ + α2 mv (109)
Consider in addition that the total vertical stress σzz is constant - then the term containing σzz in equation (108)
vanishes. From (108) and using arguments similar to the derivation of relation (106), S can be defined as
∂ζ
S= (110)
∂p ǫxx =0,ǫyy =0,σzz =c
The above expression gives S the inerpretation of being the fluid volume change per unit control volume and
per unit pressure change, while the control volume is confined in a state of zero lateral strain and constant
vertical stress.
This paragraph presents a generalisation of Terzaghi’s consolidation. We make use of equations (108) and (96).
Substituting (96) into (108) and rearranging yelds
∂p k ∂ 2p ∂σzz
S − = −αmv (111)
∂t γw ∂z 2 ∂t
This equation of mass conservation is an inhomogeneous one dimensional equation of diffusion in terms of
pore pressure, uncoupled from displacements. The right hand-side term is known (total vertical stresses are
equal to the externally applied vertical stresses), and the pore pressure can be calculated independently from
(111) without using the equations of equilibrium. If the total vertical stress is kept constant, the right-hand side
term of equation (111) vanishes. The result is Terzaghi’s diffusion equation
∂p k ∂2p
= (112)
∂t γw S ∂z 2
where the coefficient of consolidation acounts for material compressibility and is given by
23
k
c= (113)
γw S
Based on equation (101) another expression is found for the uniaxial storage coefficient, that is
1 λ + 2G
= 2 (114)
S α + (λ + 2G)Sǫ
∆p
B=− |ζ=0 (115)
∆σ
The use of pore pressure coefficient, B, as a material property was first proposed by Skempton [1954]. A general
analytical expression can be derived as follows. Substituting equation (27) in quation (93) the constitutive
relation for the increment of fluid content becomes
α α2 + KSǫ
ζ= σ+ p
K K
Making use of definition (115), the above equation yields
α
B= (116)
α2 + KSǫ
Using expression (86) for the storage coefficient Sǫ , equation (116) can be written as
C − Cs
B= (117)
C − Cs + n(β − Cφ )
In a more familiar form used in the field of soil mechanics, pore pressure coefficient can be obtained substituting
expression (87) into equation (116) as
C − Cs
B= (118)
C − Cs + n(β − Cs )
In this form pore pressure coefficient was first derived by Bishop [1973].
24
Storativity in terms of the pore pressure coefficient B
Storativity can be expressed as an equation of B and of the drained bulk modulus (or compressibility) of the
porous medium. From equation ( 116) the following expression is obtained
α (1 − αB)
Sǫ = (119)
B K
K
The term 1−αB will be defined in equation (125) as the undrained bulk modulus of the porous medium, Ku ,
which makes equation (119) equal to
α
Sǫ = (120)
BKu
σ = Ku ǫ (121)
σkk
with σ = .The stress-strain relation for a saturated medium was derived in equation (24). We also make
3
9
use of relation
α a
ζ = σ+ p (122)
K KB
and substitute for αp to (24). For undrained conditions, where ζ = 0, this results to
2G
σij = 2Gǫij + K − ǫδij + αBσδij (123)
3
We can derive the values of shear and bulk moduli from the above system of equations and the definition in
(121). First, it is easy to see from (123) that
which is the same with the case of drained deformation. This means that shear modulus is the same in drained
and undrained deformations.
The undrained bulk modulus can be derived as follows. From (123) the contracted volumetric constitutive
equation is
K
σ= ǫ (124)
1 − αB
Invoking the definition in (121) we can conclude that the undrained bulk modulus can be expressed as
K
Ku = (125)
1 − αB
Using the above equation and equation (120) a new relation for Biot’s modulus can be derived as
B 2 Ku2
M= (126)
Ku − K
9 Which is derived making use of equations (93) and (119).
25
The rest of undrained elasticity constants can be defined similar to classical elasticity theory. The undrained
Lamé constant is defined as
2
λu = Ku − G (127)
3
and it is directly equivalent to the equation for the drained Lamé constant
2
λ=K− G (128)
3
Subtracting equation (128) from (127) provides
λu − λ = Ku − K (129)
λu = λ + αBKu (130)
Another useful expression for the undrained Lamé constant can be derived from equation (129) using equations
(120) and (94) and reads
λu = λ + α2 M (131)
Similarly we can derive the following relation for the undrained bulk modulus, known as Gassman’s equation
α2
Ku = K + α2 M = K + (132)
aCs + n(β − Cφ )
The undrained Poisson’s ratio is defined as
λu
νu = (133)
2(λu + G)
A useful expression for the undrained Poisson ratio can be derived using equations (133), (130), (125), and the
third of equations (2). It expresses νu in terms of the pore pressure coefficient B as
3ν + αB(1 − 2ν)
νu = (134)
3 − αB(1 − 2ν)
Equation (134) was first derived by Rice and Cleary [1976]. Lastly, departing from the relations Gu = G and
E
G= , the undrained Young modulus is defined as
2(1 + ν)
1 + νu
Eu = E (135)
1+ν
Eu Eu νu 2(1 + νu )
Ku = ; λu = ; Ku = G (136)
3(1 − 2νu ) (1 + νu )(1 − 2νu ) 3(1 − 2νu )
26
8.3 Constitutive relations
We are now ready to derive the undrained desctiption of the constitutive relations (in terms of the increment of
fluid content and undrained elastic constants). The stress-strain relations are derived making use of equations
(24) in which pressure is substituted from equation (95), yielding
Furthermore, strain-stress relations are derived from (23) using equations (95) and (134) as
1 νu B
ǫij = σij − σkk δij + ζδij (138)
2G 1 + νu 3
αǫzz + Sǫ p = 0 (140)
′ ′
Using the one-dimensional constitutive law (ǫzz = mv σzz ) and the principle of effective stress (σzz = σzz +
αp), equation (140) leads to the reformulation
Sǫ
mv σzz = αmv + p (141)
α
From equations (139) and (141) can be concluded that the uniaxial pore pressure coefficient is equal to
mv
B′ = (142)
Sǫ
αmv +
α
B ′ can be expressed in terms of Skempton’s coefficient B by substituting for mv and Sǫ into equation (142).
To obtain this result, equation (142) is written in the form
1 Sǫ mv
= +α (143)
B′ α
10 A similar approach to the one adopted here (which leads to equation (142)) can be found in Lancellotta [2008]. Lancelotta considers
that the fluid flux is zero (−∇q = 0) and he integrates the equation of mass conservation for zero initial conditions.
27
1
0.9
ν = 0.4 →
0.8
0.7
0.6 ν = 0.5 →
B’
0.5
0.4
← ν = 0.3
0.3
0.2
← ν = 0.15
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
B
Figure 1: Relationship between Skempton’s coefficient B, and the uniaxial pore pressure coefficient, B ′ .
The coefficient of one-dimensional compressibility can be expressed in terms of the bulk modulus and Poisson’s
ratio as
1+ν
mv = (144)
3(1 − ν)K
Substituting equations (144) and (119) into equation (143) results to 11
B(1 + ν)
B′ = (146)
(1 + ν) + 2(1 − 2ν)(1 − αB)
The variation of B ′ with respect to B for various values of Poisson’s ratio is plotted in figure 1. For Poisson’s
ratio equal to ν = 1/2, the two coefficients coincide; for other values of Poisson’s ratio, B ′ is smaller than B.
There is a second approach to defining the uniaxial pore pressure coefficient B ′ , which can be found in
Section 3.6.2 of Wang [2000]. Making use of the stress-strain relation in terms of the increment of fluid content
(equation 137), for undrained behaviour (where ζ = 0) and under uniaxial strain, the conditions ǫxx = 0 and
ǫyy = 0 yield
νu
σxx = σyy = σzz (147)
1 − νu
Equations (147) lead to the mean stress being expressed as
1 + νu
σkk = σzz (148)
1 − νu
11 In the case of α = 1, equation (145) can be found in Okusa [1985].
28
From the definition of the pore pressure coefficient, B, provided in equation (115), the pore pressure applied to
the porous medium under conditions of uniaxial strain is
B(1 + νu )
3p = − σzz (149)
1 − νu
Using the definition (139), the coefficient B ′ can now be calculated from (148) as being
1 B(1 + νu )
B′ = (150)
3 (1 − νu )
For incompressible media, the undrained Poisson ratio turns out to be νu = 0.5 (see equation 157) and B ′ = B.
For highly compressible material, νu approaches zero, and B ′ ≈ B/3.
At this point it is interesting to turn the attention once again on the coefficient of consolidation. The coeffi-
cient of consolidation was expressed in equation (101) in terms of Lamé’s constants λ and G and the storativity
Sǫ . In view of the equation (33) of the one dimensional compressibility, the coefficient of consolidation can be
written with respect to mv as
k 1
c= · (151)
γw 2
Sǫ
mv α +
mv
Using equation (142), equation (151) can be further modified to take the form
k B′
c= (152)
γw m v α
Using the above equation an additional expression of the coefficient of consolidation is derived as
k G ′
c= B (154)
γw η
Finally, comparing equations (154) and (113), a new expression for the uniaxial pore pressure coefficient can
be derived, and reads
η
B′ = (155)
GS
3ν + αB(1 − 2ν)
νu =
3 − αB(1 − 2ν)
29
Another useful expression of the undrained Poisson ratio can be obtained in terms of storativity Sǫ by using
equations (134) and (119). Equation (119) can be rewritten in the form
α
B= (156)
KSǫ + α2
λSǫ + α2
νu = (157)
2(λ + G)Sǫ + 2α2
where K has been substituted for λ + 2G/3. Lastly, using Biot’s modulus instead of storativity the above
equation becomes or
λ + α2 M
νu = (158)
2(λ + G) + 2α2 M
p = M ζ − αM ǫ (160)
∂ǫ ∂ζ
G∇2 ui + λ + α2 M + G = αM − bi (161)
∂xi ∂xi
∂ǫ ∂ζ
G∇2 ui + (λu + G) = αM − bi (162)
∂xi ∂xi
Since the relation λu + G = G/(1 − 2νu ) holds, equilibrium equations can also appear as
G ∂ǫ ∂ζ
G∇2 ui + = αM − bi (163)
1 − 2νu ∂xi ∂xi
30
of porous media with incompressible constituents. Specifically, use of this formulation is made in Merxhani
[2013], where the completeness of displacement functions appropriate for consolidation problems is proven.
The pair of constitutive equations (27) - (28) is used
3ν 2Gα
σij = 2Gǫij + σδij − pδij and σ + αp = Kǫ (164)
1+ν 3K
This is similar to the pair of constitutive equations (12) for incompressible elasticity. Mean stress, σ, appearing
in them is treated as an independent variable which, however, can always be eliminated by substituting the right
hand side equation of (164) into the left hand side system of equations, resulting to the stress-strain relations
(22). From the above pair of constitutive equations the following one can be derived. When equation (122) is
substituted for αp
3νu 2GB
σij = 2Gǫij + σδij − ζδij (165)
1 + νu 3
and
3 (1 − 2νu )
ǫ= σ + Bζ (166)
2G (1 + νu )
(165) is derived from (164) making use of equations (122), (130), (129), and (127). Equation (166) can be
calculated directly from (165). Given the constitutive relation (165), the equations of equilibrium can be derived
using the same process with Section 2.2. In the absence of body forces, the equations of equilibrium in terms
of stresses are
∂σij
=0 (167)
∂xj
The stress-strain relation (165) can be used with the equilibrium equations (167) to yield
∂ 3νu ∂σ 2GB ∂ζ
G∇2 ui + G ǫ+ = (171)
∂xi 1 + νu ∂xi 3 ∂xi
with
3 (1 − 2νu )
ǫ= σ + Bζ (172)
2G (1 + νu )
For an incompressible constituent model, where Sǫ = 0 and α = 1, the undrained Poisson ratio is equal
to νu = 1/2, and the pore pressure coefficient is B = 1. Therefore, for incompressible material behaviour,
equilibrium is given by the system
∂σ 1 ∂ζ
G∇2 ui + =− G , ǫ=ζ (173)
∂xi 3 ∂xi
31
9 Variational formulation and FEM
The main numerical method used to solve consolidation problems is the finite element method. In this method,
an appropriate discretisation is applied to the weak formulation of the problem. A weak formulation for consoli-
dation problems is derived in this section, and is associated with the finite element discretisation of the problem.
The weak formulation is derived for drained description of equations with primary variables displacements and
pore pressure (u − p formulation).
The portions of the boundary Γp and Γq are the parts of the boundary in which pressure and pressure flux are
specified, with Γp ∪ Γq = Γ and
∂p
p = p̄ on Γp , and = q̄ on Γq (175)
∂n
The traction boundary condition on (174) is defined such as ti = σij ni , with n being the outer unit vector
perpendicular to the surface Γ. Using index notation, the components of the surface traction are given by the
relations
′
ti = (σij − αδij p)ni (176)
Expanding the components of eq.(176) for the two dimensional case, the following relations are obtained
′
tx = σxx nx + σxz nz = (σxx − ap)nx + σxz nz
′
(177)
tz = σzx nx + σzz nz = σzx nx + (σzz − ap)nz
From (176) is concluded that both effective stresses and pore pressure should be defined as a loading boundary
condition for the displacements.
The consolidation problem for porous linear elastic materials is defined by equations (42), (35), (41), and
(84)
∇Ts σ + b = 0
(178)
σ = σ ′ − αmp, σ ′ = D∇s u
∂ǫ ∂p k
α + Sǫ =∇· ∇p (179)
∂t ∂t γw
32
and the boundary conditions
u = ū on Γu , and t = t̄ on Γt
∂p (180)
p = p̄ on Γp , and = q̄ on Γq
∂n
Equations (178) lead to the equilibrium equations (47) or (48), for the three-dimensional or the plane strain
problem respectively, and are associated with the boundary conditions (180,1 ), while the boundary conditions
(180,2 ) are associated with the storage equation (179).
Splitting the tensor δui,j into symmetric and antisymmetric parts, the multiplication of its antisymmetric part
by σij - which is symmetric - results in zero. Furthermore using the Gauss divergence theorem and the condition
that δui = 0 on Γu , equation (182) becomes
1
Z Z Z
− (δui,j + δuj,i )σij dΩ + δui σij nj dΓ + δui bi dΩ = 0 (183)
2
Ω Γt Ω
where the symmetric gradient operator ∇s was introduced in Section 4.2, and the boundary traction, t̄, was
defined in equations (176) and (180,2 ). Finally, using equations (178,2 ), the weak form of the equilibrium
equation is obtained:
Z Z Z Z
(∇s δu)T D∇s u dΩ − (∇s δu)T amp dΩ = (δu)T t̄ dΓ + (δu)T b dΩ (185)
Ω Ω Γt Ω
Following Zienkiewicz et al. [2005], the solid skeleton displacements, u, the weighting function, δu, and the
pore fluid pressure, p, are discretised as
33
with N and Np being the shape functions of u and p and ũ and p̃ representing the values at the element nodes.
The arbitrariness of δu implies that δ ũ is arbitrary as well. Substituting (186) into (185) yields
The matrices K and Q and the vector, f , appearing in equation (188) are specified as
The weak form of the storage equation is derived next. The storage equation (84) can be written as
k
−∇· ∇p + αǫ̇ + Sǫ ṗ = 0 (190)
γw
with the upper dot denoting derivation in time. The time derivative of the volumetric strain, ǫ̇, can be expressed
as
ǫ̇ = mT ǫ̇ = mT ∇s u̇ (191)
with ǫ, m, u, and ∇s defined in Section 4.2. Multiplying (190) by an arbitrary function, δp, such that δp =
0 on Γp , and integrating over the domain results in
k
Z Z Z
− (δp)T ∇ · ∇p dΩ + (δp)T αmT ∇s u̇dΩ + (δp)T Sǫ ṗdΩ = 0 (192)
γw
Ω Ω Ω
Integrating by parts the first term of (192) and applying the divergence theorem results in the weak form of the
storage equation
k k
Z Z Z Z
T T T T
(δp) αm ∇s u̇dΩ + (δp) Sǫ ṗdΩ + (∇δp) ∇pdΩ = (δp)T q̄dΓ (193)
γw γw
Ω Ω Ω Γq
where the condition δp = 0 on Γp is used. The parameter q̄ on the right hand side term of (193) is the pore
pressure flux on the boundary Γq , defined in equation (180,2 ). Using the same discretisation technique as in
equation (185), equation (193) results in the discretised scheme
where Q was defined in equation (189). The rest of the terms that appear in equation (194) are defined as
k
S = (∇Np )T Sǫ ∇Np dΩ, NTp
R R
H= ∇Np dΩ
Ω Ω γw
k (195)
NTp
R
and q= q̄dΓ
Γq γw
34
The discrete coupled system of equations (188) and (194) can now be written as
" #" # " #" # " #
0 0 ũ˙ K −Q ũ f
+ = (196)
QT S ˙p̃ 0 H p̃ q
S≈0 (197)
H≈0 (198)
the behaviour of the consolidated system is practically undrained and the problem defined by the system (196)
is of a saddle point type (Zienkiewicz et al. [1999, 2005]). Approximating the variables of a saddle point prob-
lem with the same shape functions results in the loss of coercivity of the numerical approximation, and the
error of the approximation is no longer bounded (Ern and Guermond [2004]). The problem becomes stable
when the Babuska-Brezzi condition (Brezzi [1974]), or the equivalent patch test of Zienkiewicz et al. [1986] is
satisfied. The satisfaction of the Babuska-Brezzi condition requires the pressure field to be approximated with
polynomials of a lower degree than the polynomials approximating the displacement field. This is the reason
why different shape functions N and Np were used in (186). A compatible pair of polynomial approximation
is that of a quadratic approximation of displacements and a linear approximation of the pressure. An example
of an element satisfying the Babuska-Brezzi condition is the Taylor-Hood element, which is widely used in
modelling problems of saddle point type, for example like the ones of incompressible fluid flow.
35
10 Epilogue
The theory of linear poroelasticity is derived for isotropic fluid saturated media. The material considered is
isotropic, and undergoes quasi-static deformations under isothermal conditions. Porous space is considered to
be connected - however, the existence of isolated voids or cracks within the solid skeleton is not excluded.
Solid phase is compressible and is not necessarily composed of a single constituent. Furthermore, pore fluid is
compressible and consists of a single phase.
Both drained and undrained descriptions of poroelasticity are presented. The exposition starts with Ver-
ruijt’s drained formulation of poroelasticity - this approach uses a direct proof of equations of equilibrium and
fluid mass balance using the principle of effective stress and Darcy’s law of fluid flow in porous media. So far
pore pressure is used as an independent variable in the description. The only new coefficients that are necessar-
ily introduced in the equations are Biots coefficient that appears in the principle of effective stress, the storage
coefficient that appears in the equation of fluid mass balance, and definitions of material compressibility. In the
derivation of fluid mass balance presented, storativity (and consequently Skempton’s pore pressure coefficient)
is expressed in a more general form than in Verruijt’s original formulation, since it also includes the effect of
the unjacketed pore compressibility. The undrained description of poroelasticity is derived leading to Rice and
Cleary’s formalism. The constitutive law for the increment of fluid content as derived follows naturally from the
equation of fluid mass balance. Pore pressure coefficients, B and B ′ , and uniaxial storage coefficient, S, are de-
rived from the constitutive equation of the increment of fluid content. Other undrained poroelastic coefficients
are also introduced (Ku , νu λu , Eu ) and useful relations are proven. Where appropriate, the physical meaning
of these coefficients is proven mathematically. Lastly, a weak formulation is derived for drained description of
equations with primary variables displacements and pore pressure (u − p formulation).
Texts that can be consulted to further readers’ knowledge of poroelasticity include the following. Verruijt
[2008, 2013], who approaches poroelasticity using a soil mechanics viewpoint, and the general reviews of
Detournay and Cheng [1993] and Chau [2012]. The book of Wang [2000] that provides a complete text on the
subject with applications to geomechanics and hydrogeology. Furthermore, Coussy [2004] offers an advanced
treatment of poromechanics, where the theory is systematically derived departing from a general formulation
in terms of large deformations. In the above citations a plethora of examples is also included.
References
M. Biot. General theory of three dimensional consolidation. Jurnal of Applied Physics, 12:155–164, 1941.
M. Biot. Theory of elasticity and consolidation for a porous anisotropic solid. Journal of Applied Physics, 26
(2):182–185, Feb. 1955.
M. Biot. Theory of propagation of elastic waves in fluid-satureted porous solid. Journal of the Acoustical
Society of America, 28:168–178, 1956.
M. Biot. Nonlinear and semilinear rheology of porous solids. Journal of Geophysical Research, 78:4924–4937,
1973.
M. Biot and D. G. Willis. The elastic coefficients of the theory of consolidation. Journal of Applied Mechanics,
pages 594–601, 1957.
36
A. W. Bishop. The influence of an undrained change in stress on the pore pressure in porous media of low
compressibility. Geotechnique, 23:435–442, 1973.
F. Brezzi. On the existence uniqueness and approximation of saddle-point problems arising from lagrangian
multipliers. R.A.I.R.O. Analyse Numerique, 8, R-2:129–151, 1974.
P. Chou and N. Pagano. Elasticity: Tensor, Dyadic, and Engineering Approaches. Dover Books on Engineering.
Dover Publications, 1992.
O. Coussy. Méchanique des milieux poreux. Éditions Technip, 1991. ISBN 2710805952.
R. de Boer. Highlights in the historical development of the porous media theory: Toward a consistent macro-
scopic theory. Applied Mechanics Reviews, 49:201–262, 1996.
R. de Boer. Theory of Porous Media: Highlights in Historical Development and Current State. Springer, 2000.
A. Ern and J. Guermond. Theory and practice of finite elements. Applied mathematical sciences. Springer,
2004.
R. Lancellotta. Geotechnical Engineering 2E. Spon Text. Taylor & Francis, 2008.
A. Merxhani. Analytical and Numerical Study of Poroelastic Wave-Seabed Interactions. PhD thesis, Jesus
College, University of Cambridge, 2013.
J. Mitchell and K. Soga. Fundamentals of soil behavior. John Wiley & Sons, 2005.
S. Okusa. Wave-induced stress in unsaturated submarine sediments. Geotechnique, 35, no 4:517–532, 1985.
J. H. Prevost. Mechanics of continuous porous media. International Journal of Engineering Science, 18:
787–800, 1980.
J. H. Prevost. Nonlinear transient phenomena in saturated porous media. Computer Methods in Applied
Mechanics and Engineering, 20:3–8, 1982.
J. Rice and M. Cleary. Some basic stress diffusion solutions for fluid-saturated elastic porous media with
compressible constituents. Reviews of Geophysics and Space Physics, 14(2), May 1976.
J. W. Rudnicki. Fluid mass sources and point forces in linear elastic diffusive solids. Mechanics of Materials,
5:383–393, 1986.
J. W. Rudnicki. Fundamentals of continuum mechanics. John Wiley & Sons, 2015. ISBN 9781118479919.
A. W. Skempton. Effective stress in soils, concrete, and rock. In Pore pressure and suction in soils. Butter-
worths, London, 1960.
37
K. Terzaghi. Erdbaumechanik auf bodenphysikalischer Grundlage. Deuticke, Wien, 1925.
A. Verruijt. Elastic storage of aquifers. In R. D. Wiest, editor, Flow through porous media. 1969.
A. Verruijt. Consolidation of soils. In Encyclopedia of Hydrological Sciences. John Wiley & Sons, 2008.
H. F. Wang. Theory of linear poroelasticity, with applications to geomechanics and hydrogeology. Princeton
University Press, 2000.
H. Westergaard. Theory of elasticity and plasticity. Harvard monographs in applied science. Harvard University
Press, 1952.
O. Zienkiewicz, R. Taylor, and J. Zhu. The finite element method: its basis and fundamentals. The Finite
Element Method. Elsevier Butterworth-Heinemann, 2005.
O. C. Zienkiewicz, C. T. Chang, and P. Bettess. Drained, undrained, consolidating and dynamic behaviour
assumptions in soils. limits and validityy. Geotechnique, 30:385–395, 1980.
O. C. Zienkiewicz, S. Qu, R. L. Taylor, and S. Nakazawa. The patch test for mixed formulations. International
Journal for Numerical Methods in Engineering, 26:1039–1055, 1986.
38